US20010014140A1

US20010014140A1 - Method of combining reconstruction images

Info

Publication number: US20010014140A1
Application number: US09/732,587
Authority: US
Inventors: Roland Proksa; Michael Grab; Thomas Kohler
Original assignee: US Philips Corp
Current assignee: US Philips Corp
Priority date: 1999-12-08
Filing date: 2000-12-08
Publication date: 2001-08-16
Anticipated expiration: 2020-12-08
Also published as: DE19959092A1; US6381298B2; GB2361613A; GB0029951D0; JP2001198120A; GB2361613B

Abstract

The invention relates to a method of and an X-ray device for forming a 3D image (B) of an object (13) to be examined by combining at least two reconstruction images (S₁, S₂) by weighted addition, said reconstruction images being acquired notably by means of an X-ray device (1) and each reconstruction image (S₁, S₂) being weighted by a respective weighting function (A₁, A₂) that describes at least approximately the distribution of noise and/or artefacts in the relevant reconstruction image. The image quality of the resultant 3D image is thus significantly enhanced.

Description

The invention relates to a method of forming a 3D image of an object to be examined by combining at least two reconstruction images. The invention also relates to an X-ray device which is particularly suitable for carrying out such a method.

It is known to utilize two-dimensional projections of an object to be examined for the formation of a reconstruction image of the object to be examined; such projections have been acquired by means of an imaging system, for example an X-ray device such as a C-arm X-ray device, a computed tomography apparatus, a magnetic resonance tomography apparatus or an ultrasound device. Various reconstruction methods that utilize different algorithms, for example, the Feldkamp algorithm, are known for the reconstruction. They enable the formation of a reconstruction image of an object to be examined from two-dimensional projections acquired along a circular trajectory of the measuring device, that is, for example the X-ray source and the X-ray detector, around the object to be examined. However, this type of data acquisition usually does not provide adequate data for the inverse problem, that is, the reconstruction of a complete three-dimensional reconstruction image; this becomes apparent notably in the so-called Radon space. For example, a data acquisition along a circular trajectory provides data only within a torus in the Radon space; this does not suffice for the formation of an exact three-dimensional reconstruction image for which data would be required within a complete sphere in the Radon space. Therefore, the Feldkamp algorithm is merely an approximation and yields a reconstruction image which is exact in the central layer whereas induced artefacts increase continuously as the distance from the central layer increases.

Granted, it is possible to mitigate this problem by combining data acquired along two or more trajectories. For example, from EP 860 696 A2 it is known to acquire projections along two semi-circular trajectories that extend at an angle of 60° relative to one another, to form a respective reconstruction image from the projections acquired along each time one trajectory, and to add the two reconstruction images subsequently so as to form a 3D image. The image quality can thus be improved, but artefacts still occur in the 3D image.

Therefore, it is an object of the invention to provide a method and an X-ray device for forming a 3D image of an object to be examined while improving the image quality.

This object is achieved by means of a method as disclosed in

claim

1 and by means of an X-ray device as disclosed in claim 8.

The invention is based on the recognition of the fact that noise and artefacts occurring in the reconstruction image are transferred to the resultant 3D image when two or more reconstruction images are simply added. The noise is due essentially to the hardware used, notably the detector elements used, for example the X-ray detector, whereas the cause of the artefacts lies mainly in the fact that not all data necessary for an exact reconstruction can be acquired from projections acquired along a trajectory. Depending on the reconstruction algorithm used, a different type of noise and different artefacts are thus induced in a reconstruction image. In order to reduce the transfer of such noise or such artefacts from a reconstruction image to a resultant 3D image, the invention proposes to determine a weighting function for each reconstruction image, which weighting function at least approximately describes the distribution of noise and/or artefacts in said reconstruction image, to multiply the reconstruction image by said weighting function, and to add all reconstruction images thus weighted so as to form a resultant 3D image only after that.

Advantageous versions of the method in accordance with the invention and the X-ray device in accordance with the invention are disclosed in the dependent claims.

In preferred further versions of the method in accordance with the invention the weighting functions are determined by simulations or measurements performed on a phantom object, or use is made of mathematical functions, for example functions which descend linearly or as a square root (or vary otherwise), for example, from the center to the edge of the reconstruction image to be weighted by the respective weighting function. The image quality of the resultant 3D image can be significantly improved in comparison with the known method even while using such simple mathematical functions as weighting functions.

Further improvements still can be achieved when the weighting function is adapted as accurately as possible to the distribution of the noise or the distribution of artefacts occurring in a reconstruction image. It may be arranged notably that to each pixel or group of pixels of a reconstruction image there is assigned a value of the associated weighting function, so that each pixel or each group of pixels can be individually weighted in dependence on the magnitude of the noise component or artefact component of this pixel or this group of pixels.

In a particularly attractive version of the method the weighting functions are chosen in such a manner that artefacts and noise are separately weighted for each pixel or each group of pixels of a reconstruction image. This enables advantageous selection of the preferred type of disturbance and the location in the 3D image in which such disturbances are to be suppressed.

The invention is used particularly advantageously in an X-ray device that is provided with an X-ray source that generates a conical X-ray beam and with a two-dimensional X-ray detector, the X-ray source and the X-ray detector rotating about the object to be examined in order to acquire the projections. The X-ray device may be, for example, a C-arm X-ray unit. In principle, however, the invention can also be used in a computed tomography apparatus.

The invention will be described in detail hereinafter with reference to the drawings. Therein: [0012]
FIG. 1 shows a C-arm X-ray unit in accordance with the invention, [0013]
FIG. 2 shows diagrammatically two trajectories followed during the acquisition of projections, [0014]
FIG. 3 shows a flow chart illustrating the method in accordance with the invention, and [0015]
FIG. 4 shows a computed tomography unit in accordance with the invention. [0016]
The C-[0017] arm X-ray unit 1 shown in FIG. 1 includes an X-ray source 2 which is arranged at one end of the C-arm and an X-ray detector 3 which is arranged at the other end of the C-arm 20. The X-ray tube generates a conical X-ray beam 14 which traverses an object 13 to be examined, for example a patient, arranged on a patient table 4 in the examination zone, after which it is incident on the two-dimensional X-ray detector 3. In the position shown the X-ray tube 2 and the X-ray detector 3 are rotatable about the y axis via rails 7 provided on the C-arm 20. As a result of the suspension by way of several arms and hinges 5, 6, the position of the C-arm 20 can be changed in various directions; for example, the C-arm 20 is capable of rotation about the x axis and the z axis. The control of these motions for the acquisition of projections from different X-ray positions and of the data acquisition is performed by means of a control unit 8. The projections acquired are applied to a reconstruction unit 9 which forms a respective reconstruction image from the projections acquired along a trajectory. These reconstruction images, formed each time from a respective set of projections acquired along different trajectories, are subsequently applied to an arithmetic unit 10 which, in conformity with the method in accordance with the invention, divides each reconstruction image by a weighting function stored in a storage unit 11, the result subsequently being added and normalized. The arithmetic unit 10 is also controlled by the control unit 8. The resultant 3D image can be displayed on a monitor 12.
FIG. 2 shows diagrammatically two trajectories T[0018] ₁and T₂. Each trajectory describes the path traveled by the center of the detector surface of the X-ray detector 3 during the acquisition of projections. Therefore, the trajectory is the curve through all X-ray positions in which a respective projection has been acquired. In the case shown the trajectories T₁and T₂describe a respective semi-circle and are tilted through an angle of 2α=90° relative to one another. A first reconstruction image is formed from the projections acquired along the trajectory T₁whereas a second reconstruction image is formed from the projections acquired along the trajectory T₂. Subsequently, the two reconstruction images are combined so as to form a resultant 3D image in accordance with the invention; this will be described in detail hereinafter with reference to FIG. 3.
The [0019] blocks 201 and 202 of the flow chart shown in FIG. 3 symbolically contain two sets of projections P₁(α₁) and P₂(α₂) as starting points determined along two trajectories T₁and T₂, respectively, that extend at the angles α₁and α₂, respectively, relative to a reference plane. From each of these sets of projections P₁, P₂a reconstruction image S₁, S₂is determined in the blocks 211 and 212. In the blocks 231 and 232 the reconstruction images S₁and S₂are divided by the weighting functions A₁(α₁) and A₂(α₂), adapted to the relevant reconstruction image, or multiplied by the inverse weighting functions 1/A₁(α₁) and 1/A₂(α₂). The weighting functions A₁, A₂are stored in the blocks 221 and 222 and are determined in advance, for example, by means of a simulation calculation performed on a phantom object. Finally, in the block 24 the reconstruction images thus weighted are combined by addition and normalization so as to form one 3D image B which, depending on the quality of the weighting functions, exhibits significantly less noise and artefacts than a 3D image formed by simple addition of two non-weighted reconstruction images.
The formation of the 3D image B can also be described in general in the form of a formule for the case involving combination of n reconstruction images S[0020] _n: $B = \frac{\sum_{n} S_{n} A_{n}^{- 1}}{\sum_{n} A_{n}^{- 1}}$
It may also be arranged that a distinction is made between noise and artefacts during the weighting by means of the weighting functions; the sum of these two subweighting operations then has to be 1 for each pixel. This can also be expressed in the form of a formule for a general case: [0021] $B = \frac{\sum_{n} S_{n} (w_{n} R_{n}^{- 1} + (1 - w_{n}) K_{n}^{- 1})}{\sum_{n} w_{n} R_{n}^{- 1} + (1 - w_{n}) K_{n}^{- 1}}$
Therein: [0022]
K[0023] _nis the artefact distribution function for the reconstruction image n,
R[0024] _nis the noise distribution function for the reconstruction image n, and
w[0025] _nis the weight distribution that weights the suppression of noise and artefacts in a reconstruction image, that is, enables a decision to be taken as regards the ratio of suppression of artefacts to suppression of noise in a reconstruction image.
The method in accordance with the invention enables a significant improvement in respect of the suppression of noise and of artefacts in a 3D image. Generally speaking, the improvement of the signal-to-noise ratio that can be theoretically achieved by addition of n reconstruction images amounts to {square root}n in comparison with the signal-to-noise ratio of a single reconstruction image, provided that the noise component in all n reconstruction images is approximately the same; however, this is only rarely the case in practice. In practice, therefore, this improvement by the known addition of reconstruction images is generally significantly less. Assuming that artefacts in different reconstruction images are not correlated, the method in accordance with the invention, however, enables an improved signal-to-noise ratio to be obtained which may amount to as much as {square root}n, depending on the quality of the weighting functions. [0026]
FIG. 4 shows a computed [0027] tomography unit 17 in accordance with the invention. The X-ray device 2′ includes a collimator 19 for forming a conical X-ray beam 15 that is mounted, together with the X-ray detector 3′, on a ring-shaped gantry 18; they rotate about the object 13 to be examined, arranged along the z axis, in order to acquire projections; to this end, the gantry is controlled by a motor drive 16 which itself is controlled by the control unit 8′. The projections acquired are applied again to a reconstruction unit 19 in order to form reconstruction images which are applied again to the arithmetic unit 10. The 3D image is formed from said reconstruction images in the same way as described above for the C-arm X-ray unit.
The X-ray devices shown constitute merely exemplary embodiments of the invention. The invention, however, can also be used in other X-ray devices in which a 3D image that has been enhanced in respect of suppression of noise and artefacts is to be formed from a plurality of reconstruction images. The trajectories shown in FIG. 2 and the number of such trajectories are also given merely by way of example. The projections can also be acquired along other trajectories and also along more than two trajectories, for example two or more parallel full circles or two full circles extending perpendicularly to one another. [0028]

Claims

1. A method of forming a 3D image (B) of an object (13) to be examined by combining at least two reconstruction images (S₁, S₂) by weighted addition, each reconstruction image (S₁, S₂) being weighted with a weighting function (A₁, A₂) which describes at least approximately the distribution of noise and/or artefacts in said reconstruction image.

2. A method as claimed in

claim 1

,

characterized in that the weighting function (A₁, A₂) used is a function which describes at least approximately the distribution of noise and/or artefacts which occurs in the reconstruction image (S₁, S₂) to be weighted thereby and is due to the reconstruction method used.

3. A method as claimed in

claim 1

,

characterized in that the weighting functions (A₁, A₂) are determined by simulation or measurement performed on a phantom object.

4. A method as claimed in

claim 1

,

characterized in that use is made of weighting functions (A₁, A₂) in the form of mathematical functions, for example functions that descend linearly or as a square root from the center to the edge of the reconstruction image (S₁, S₂) to be weighted.

5. A method as claimed in

claim 1

,

characterized in that the weighting functions (A₁, A₂) are chosen in such a manner that artefacts and noise are separately weighted for each pixel of a reconstruction image (S₁, S₂).

6. A method as claimed in

claim 1

,

characterized in that the reconstruction images (S₁, S₂) are acquired by means of an imaging medical system, notably by means of an X-ray device, an ultrasound device, a magnetic resonance tomography apparatus or a computed tomography apparatus.

7. A method as claimed in

claim 1

,

characterized in that the reconstruction images (S₁, S₂) are derived from a plurality of projections (P₁, P₂) acquired by means of an X-ray device (1) which includes a radiation source (2) that generates a conical X-ray beam (14) and rotates about the object (13) to be examined, and also includes an X-ray detector (3) that rotates about the object (13) to be examined.

8. An X-ray device, notably a device for carrying out the method claimed in

claim 1

, including an X-ray source (2) and an X-ray detector (3) that are rotatable about the object (13) to be examined in order to acquire projections (P₁, P₂) from different X-ray positions, and also including a reconstruction unit (9) for forming reconstruction images (S₁, S₂) from each time a respective set of projections (P₁, P₂), the individual sets of projections (P₁, P₂) being acquired along different trajectories (T₁, T₂),

characterized in that there is provided an arithmetic unit (10) for forming a 3D image (B) of the object (13) to be examined by combining at least two reconstruction images (S₁, S₂) by weighted addition, each reconstruction image (S₁, S₂) being weighted by a weighting function (A₁, A₂) that describes at least approximately the distribution of noise and/or artefacts in said reconstruction image (S₁, S₂).

9. An X-ray device as claimed in

claim 8

,

characterized in that the X-ray source (2) and the X-ray detector (3) are arranged to produce and detect, respectively, a conical X-ray beam (14).

10. An X-ray device as claimed in

claim 8

,

characterized in that the X-ray device is a C-arm X-ray unit (1) or a computed tomography unit (17).